Front. Environ. Sci.Frontiers in Environmental ScienceFront. Environ. Sci.2296-665XFrontiers Media S.A.162632610.3389/fenvs.2025.1626326Environmental ScienceOriginal ResearchResearch on the impact of the digital economy on carbon-neutral technology innovationGeng et al.10.3389/fenvs.2025.1626326GengYingying1LiChaoyang2ChenYiming2*1Business School, Shandong University of Technology, Zibo, China2School of Economics and Management, Weifang University of Science and Technology, Weifang, China
Edited by:Katundu Imasiku, Georgia Institute of Technology, United States
Reviewed by:Rui Huang, Nanjing Normal University, China
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
At the historical intersection of the “dual carbon” strategy and the digital economy, leveraging digital power to promote environmental governance and technology innovation has emerged as a key area of study. Consequently, investigating how the digital economy influences carbon-neutral technology innovation has become a prominent area of focus. Utilizing panel data from 264 Chinese cities spanning 2011 to 2022, this study explores the influence of the digital economy and its internal structure on carbon-neutral technology innovation. The results show: (1) The digital economy possesses a potent promotional role in driving carbon-neutral technology innovation, and there is regional heterogeneity. (2) Digital industrialization and industrial digitalization, two major systems within the digital economy, have significant promoting impacts on carbon-neutral technology innovation, and compared with industrial digitalization, the promoting effect of digital industrialization is stronger. (3) The digital economy can enhance carbon-neutral technology innovation by improving resource mismatch. (4) When using the digital economy, digital industrialization, and industrial digitalization as threshold variables, the digital economy produces a nonlinear influence on carbon-neutral technology innovation. (5) The digital economy exerts a spatial spillover influence on carbon-neutral technology innovation. The research’s conclusions have certain referential value for promoting China’s digital economy and carbon-neutral technology innovation.
digital economycarbon-neutral technology innovationdynamic threshold modelspatial spillover effectregional heterogeneitysection-at-acceptanceEnvironmental Economics and Management1 Introduction
The 2024 government work report proposed to “promote comprehensive ecological and environmental governance,” “energetically foster a green and low-carbon economy,” “actively and prudently push forward carbon peaking and carbon neutrality,” “accelerate the green transformation of development models,” and “pursue green and low-carbon development.”1 In 2022, nine departments, including the Ministry of Science and Technology, jointly released the Implementation Plan for Peaking Carbon Neutrality by Science and Technology (2022–2030)2 and put forward a range of action plans from fundamental research, technology research and development (R&D), application demonstration, result popularization, talent cultivation, international collaborations, and other areas to expedite the advancement of green and low-carbon technology innovation. Against this backdrop, the fundamental way to achieve “carbon neutrality” lies in accelerating the setup of a carbon-neutral technology innovation (c_innovation) system that includes the development and deployment of climate change mitigation technologies like carbon capture, carbon storage, energy generation, and transportation, promoting carbon emission reduction and offsetting, while overcoming obstacles to low-carbon innovation.
The digital economy (d_economy) functions as a pivotal driver for the economic growth of nations worldwide (Lu et al., 2025). Its characteristics of penetration, substitution, and synergy (Han D. et al., 2025) not only provide the material and technological foundation for technology innovation but also offer a new digital foundation for environmental governance, becoming an important driving force for enabling c_innovation. During the historical convergence period of the d_economy and the “carbon neutrality” strategy, facing the complex environment where “energy constraints” are shifting to “carbon emission reduction constraints” filled with uncertainties such as the over-emphasis on heavy industries in economic structure, it carries substantial theoretical meaning and practical worth to thoroughly examine the changes in c_innovation in the digital revolution and explore how to effectively unlock the propelling force of the d_economy for c_innovation.
Academic studies on the d_economy have primarily focused on conceptual definitions (Bowman, 1996; Tapscott, 1996; Lane, 1999), measurement methods, and their impact effects. Studies on measuring the d_economy can generally be classified into two perspectives. One is from a quantitative perspective, where the scale value of the d_economy is calculated (Xie and Zhang, 2024). The other is from a qualitative perspective, where an all-encompassing evaluation index framework for the d_economy is constructed that is derived from the conceptual connotation of the d_economy (Shi et al., 2023; Yuan, 2025; Lv et al., 2025). Some scholars have constructed an assessment index system for the d_economy from dimensions such as the Internet adoption rates, the count of Internet professionals, Internet-associated outputs, and mobile Internet user numbers (Shi et al., 2023; Yuan, 2025). Others have built an index system based on four areas: digital infrastructure, digital innovation capabilities, digital industry growth, and digital financial elements (Lv et al., 2025). Because the d_economy represents an economic system based on digital technology, its connotation can be divided into digital industrialization (digital_i) and industrial digitalization (industrial_d). Some scholars have also measured it through the lenses of digital_i and industrial_d (Xue et al., 2022).
The d_economy system is extremely complex. Given the substantial disparities in the global development process of the d_economy, the academic community lacks a uniform or universal standard for its assessment. The perspectives and dimensions of research results also vary. However, developing an assessment index system has emerged as the predominant methodology employed within academic and governmental spheres for evaluating the development of the d_economy (Shi et al., 2023). Regarding impact effects, existing studies indicate that the d_economy holds a vital position in optimizing the industrial structure (Tan et al., 2024), boosting energy efficiency (Wang and Shao, 2023), and driving the green transition of industries (Yang et al., 2024). Notably, the specific mechanism by which the d_economy influences c_innovation has not yet been revealed. However, the relevant literature on how the d_economy affects green technology innovation provides a theoretical foundation for this study. The research reveals that the intrinsic nature of the d_economy, along with its high innovation capacity, powerful penetration, extensive scope, as well as its development trends and patterns, can directly influence green technology innovation, and also promote the improvement of green technology innovation levels in adjacent regions (Wang et al., 2022; Song et al., 2024). Some research has also examined the pathways through which digital elements such as big data, the Internet, and information technology contribute to green technology innovation and innovation development (Jin et al., 2021), as well as the heterogeneous impacts resulting from differences in regional scale and resource endowment (Song et al., 2019; Ghasemaghaei and Calic, 2020).
By summarizing and analyzing previous literature, it becomes apparent that the existing research on how the d_economy affects c_innovation remains in its infancy, and there are still some shortcomings: (1) Studies exploring how the d_economy affects c_innovation remain relatively scarce, and such studies lack rigorous empirical proof to support them. (2) Most current studies focus on the impact of the d_economy but often overlook the influences generated by the two major subsystems within the d_economy, namely, the digital_i and the industrial_d. The impact effects of these two internal subsystems of the d_economy still need to be further explored.
Considering this, this article chooses panel datasets from 264 Chinese prefecture-level cities spanning the years 2011 to 2022 as its research sample. Grounded in the connotation scope of the d_economy, an assessment index system for its development is constructed from two perspectives: digital_i and industrial_d. The projection pursuit method optimized by the accelerated genetic algorithm (RAGA-PP) is adopted to measure it. The article seeks to examine how the d_economy and its two major subsystems affect c_innovation by exploring the four “mirrors” that reveal the essence of things. First, from the perspective of a “flat mirror,” the fixed-effect model is applied to describe the effect of d_economy directly affecting c_innovation and explore the regional heterogeneity of d_economy on c_innovation. Second, from the perspective of a “magnifying glass,” the d_economy is divided into two internal systems, digital_i, and industrial_d, and the difference between the two on c_innovation is explored. The indirect influence of d_economy on c_innovation is explored based on the mediating effect model from the “microscope” standpoint. Using the dynamic threshold model, the nonlinear influence of d_economy on c_innovation is discussed respectively under the constraints of digital_i and industrial_d. Finally, adopting a “telescope” perspective, the spatial Durbin model is employed to further probe the spatial impacts of d_economy on c_innovation in geographical proximity, with the aim of providing a certain theoretical basis and experience reference for China’s c_innovation and regional d_economy green development.
In comparison to earlier studies, the possible marginal contributions of this article are listed below: (1) With the d_economy as the entry point, this study employs fixed-effect models, mediating effect models, dynamic threshold models, and spatial econometric models to explore its impact on c_innovation. It provides certain theoretical support and empirical proof for investigations in related fields. (2) Rooted in the connotation and scope of the d_economy, this article further divides it into digital_i and industrial_d, and explores the impact effects of the two major subsystems within the d_economy, thereby enriching the existing studies.
After the introduction, the article is structured as detailed below: The second section elaborates on this article’s mechanism analysis and research hypothesis. The third section describes the model’s creation, measurement of related variables, and the data sources. The fourth section analyzes the benchmark regression results of this article. The fifth section further analyzes this study’s empirical findings. Finally, we draw the main conclusions and recommendations.
2 Mechanism analysis and research hypothesis2.1 Analysis and research hypothesis on the impact of <inline-formula id="inf65">
<mml:math id="m65">
<mml:mrow>
<mml:mi mathvariant="italic">d</mml:mi>
<mml:mo>_</mml:mo>
<mml:mi mathvariant="italic">e</mml:mi>
<mml:mi mathvariant="italic">c</mml:mi>
<mml:mi mathvariant="italic">o</mml:mi>
<mml:mi mathvariant="italic">n</mml:mi>
<mml:mi mathvariant="italic">o</mml:mi>
<mml:mi mathvariant="italic">m</mml:mi>
<mml:mi mathvariant="italic">y</mml:mi>
</mml:mrow>
</mml:math>
</inline-formula> on <inline-formula id="inf66">
<mml:math id="m66">
<mml:mrow>
<mml:mi mathvariant="italic">c</mml:mi>
<mml:mo>_</mml:mo>
<mml:mi mathvariant="italic">i</mml:mi>
<mml:mi mathvariant="italic">n</mml:mi>
<mml:mi mathvariant="italic">n</mml:mi>
<mml:mi mathvariant="italic">o</mml:mi>
<mml:mi mathvariant="italic">v</mml:mi>
<mml:mi mathvariant="italic">a</mml:mi>
<mml:mi mathvariant="italic">t</mml:mi>
<mml:mi mathvariant="italic">i</mml:mi>
<mml:mi mathvariant="italic">o</mml:mi>
<mml:mi mathvariant="italic">n</mml:mi>
</mml:mrow>
</mml:math>
</inline-formula>
The d_economy, with its characteristics of high penetration, speed, and increasing marginal effect, is prompting significant transformations in production, lifestyles, and governance and playing an essential part in reducing urban carbon emissions (Wang L. et al., 2024; Wang Y. et al., 2024). Achieving the carbon peaking and carbon neutrality strategy requires green technology innovation, and low-carbon technology and innovations in green technology are critical to enterprise conservation of energy and emission reduction. Reducing coal use, increasing energy efficiency, and creating energy from renewable sources are three pivotal issues in reaching the carbon peak by 2030. All of these demand backing from technological “underpinning,” particularly the advancement of carbon-neutral and green technology innovations.
The continuous integration of the d_economy with green technology innovation in businesses, universities, and academic institutes can markedly spur urban green technology innovation and promote c_innovation. The d_economy promotes c_innovation in the following three areas: First, regarding human capital, the d_economy has spawned numerous emerging industries rooted in digital technology, including big data and blockchain, thereby drawing in a flow of high-caliber talents. Digital technology’s broad-based adoption and application will boost the requirement for highly skilled and well-educated workers (Han J. et al., 2025), hence continually enhancing the structure of human capital. The refinement of the human capital structure lays a robust groundwork of innovative elements for the growth of urban c_innovation, and contributes to elevating the standard of urban carbon-neutral and green technology innovation (Ling et al., 2024). Second, regarding financing constraints, financing restrictions and funding availability exert a significant influence on enterprise innovation and green technology innovation (Hall et al., 2016; Cao et al., 2021). The platform effect of the d_economy can transcend time and space constraints, strengthen the ability to process information, ease the information asymmetry problem between banks and companies, and empower financial institutions to precisely assess enterprise operations. This allows for efficient provision of credit funds to support enterprise development and optimizes the deployment of bank resources. Moreover, the relaxation of corporate financing constraints enables more funds for green technology innovation and development, thus boosting the advancement and innovation of green technologies (Cao et al., 2021). Decreased financing constraints can encourage economic entities to expand R&D spending and introduce R&D personnel to carry out c_innovation. Third, in terms of industry–university–research cooperation, the use of digital technology can remove obstacles to the flow of information (Zhang et al., 2025). It not only allows companies to promptly grasp the market needs for low-carbon technologies and products but also reinforces collaboration and links between enterprises, universities, and research institutions, enhances the collaborative innovation capabilities of enterprises, continuously improves the standard of industry–university–research collaboration, and promotes the improvement of the c_innovation level.
It is worth noting that the Metcalfe Law (Zhao et al., 2020) and the existence of “network effects” can potentially empower the d_economy to exert a marginal incremental influence on c_innovation. As the d_economy advances in development, data element resources are no longer scarce, and the promoting role of c_innovation caused by the high penetration and fast characteristics of the d_economy is also enhanced.
In light of this, this research suggests Hypothesis 1: d_economy exerts a positive boosting influence on c_innovation. It also shows the effect of marginal increase, meaning that as the d_economy advances in development, its capacity to foster creative c_innovation gradually intensifies.
digital_i and industrial_d are two primary parts of the d_economy, and their simultaneous advancement is critical to fostering economic transformation and upgrading. In realizing and releasing data element value, digital_i and industrial_d play key roles.
digital_i, as the provider of digital technologies and data elements, can offer necessary digital technologies, products, and solutions such as the Internet of Things, big data, and cloud computing to the digitalization, networking, and intelligent transformation of traditional industries (Wang and Qi, 2023). The technological transformation led by digital_i has stimulated the development potential of new technologies. Existing studies have shown that digital_i contributes positively to enhancing innovation capabilities and upgrading the industrial structure (Luo et al., 2023; Sturgeon, 2019), and the boosting of innovation capabilities and the refinement of the industrial structure are essential approaches to reduce carbon emissions and boost c_innovation. Improving innovation capacity can promote negative-carbon technologies like carbon capture and disposal, reduce energy usage and pollutants, and eventually lower carbon emissions while promoting c_innovation.
During the initial phase of digital_i, due to underdeveloped related infrastructure, a scarcity of skilled personnel, and insufficient exploration of digital technology application scenarios, it is challenging for the d_economy to fully exert its influence in integrating with traditional industries, and its role in promoting c_innovation is limited. As digital_i advances, the powerful economies of scale effect of the d_economy becomes prominent, drawing substantial capital towards investments in R&D for carbon-neutral technologies, optimizing resource allocation, promoting knowledge and technology spillover, accelerating the agglomeration and integration of innovation elements, and thus significantly increasing the c_innovation level.
industrial_d is the core of the d_economy and the process of increasing output and improving efficiency caused by the application of data elements, digital technologies, and digital intelligence products in traditional real industries. industrial_d relies on blockchain and other technologies to enable green development and promote c_innovation, with this mainly reflected in two aspects: On the one hand, industrial_d can promote inter-regional linkages of industries, form an innovation ecosystem with rapid flow of data, talent, and capital, and improve regional core technological innovation capabilities (Wang and Qi, 2023). On the other hand, industrial_d reorganizes the industrial competition model and promotes the integration of industrial boundaries. Based on the theory of industrial integration, it helps to accelerate the integration of internal resources of enterprises, realize resource sharing, promote R&D capabilities, thus encouraging enterprises to engage in more innovative activities, and promote enterprises to R&D non-carbon technologies with lower carbon emission levels and zero-carbon emission, as well as negative-carbon technologies that compensate for process-related emissions. This effectively empowers green and low-carbon industrial development, improves carbon productivity, and lays a foundation for c_innovation.
At lower stages of industrial_d, the communication channels of digital technology among various industries are not smooth, the information barriers between different industries are high, the d_economy is difficult to effectively integrate resources, and its role in promoting c_innovation is limited. As industrial_d advances further, the d_economy expands in scale, the collaborative innovation between industries increases, and the technology and knowledge exchange and integration of different industries have accelerated the process of c_innovation.
Notably, digital_i refers to the process where digital technologies continuously innovate and their market applications expand, thereby forming a digital industry with characteristics such as high penetration, technology-intensive nature, and foundational nature. digital_i, as the industrialization process of digital technology itself, represents a breakthrough from 0 to 1. It has the characteristics of rapid technological iteration and strong innovation spillover effects. It can leverage the advantages of digital technology to drive traditional industries towards intelligent and green development, providing foundational support for c_innovation. Meanwhile, the introduction and implementation of a suite of policies like the “14th Five-Year Plan” have created a conducive setting for digital_i to exert its empowering role and for the advancement of c_innovation. industrial_d refers to the process of integrating digital technologies with the real economy (Xue et al., 2022). Currently, China’s industrial structure still faces the problems of being “big without being powerful” and “comprehensive but not excellent.” Traditional high-energy-consuming industries make up a relatively large share, and the integration of traditional industries and digital industries requires a process, which leads to a relatively slow progress of industrial_d. The promoting effect of this on c_innovation has not yet been fully manifested.
With regard to this, this article suggests Hypothesis 2: digital_i and industrial_d can promote the improvement of c_innovation, showing the characteristics of “marginal increase.” Compared with digital_i, industrial_d serves a more significant role in encouraging c_innovation.
The development of the d_economy can improve resource allocation efficiency and address issues of mismatched resources, thereby generating a favorable influence on c_innovation. The d_economy, leveraging technologies like cloud computing and big data, reorganizes resource allocation and alleviates the distortions in the factor market allocation (Chen, 2020; Xu et al., 2022), achieving precise matching of factors through penetration and synergy, enhancing resource allocation efficiency, and correcting the problem of excessive resource allocation. In addition, the d_economy promotes organizational innovation, breaks down regional and industry barriers, and optimizes investment efficiency and customer channels, thereby improving resource allocation efficiency. Improving resource mismatch can effectively integrate technology resources among different industries, promote technology sharing and transfer, and improve the advancement of c_innovation.
In light of this, the article proposes Hypothesis 3: The d_economy can improve c_innovation by improving resource mismatch.
Amid the context of the d_economy, digital information has become a key new production factor, and information technology has developed into a significant carrier for driving economic and efficient operations. In the digital network, the flow of information can break through geographical constraints and overcome space and industry barriers, exerting the superimposed effect of “flow space” and “flow industry.” Simultaneously, based on the sharing and penetration characteristics of the d_economy, key resources dominated by technology innovation and knowledge realize cross-regional flow. This means the impact of the d_economy is not confined to a single region. Studies demonstrates that China’s d_economy exhibits considerable spatial spillover, especially in promoting innovation and economic growth in surrounding urban areas (Huang et al., 2022). Relying on the spatial correlation between social and economic growth, the economy’s ability to encourage technology innovation is likely to be spatially correlated. The information flow and technology spillover across spatial constraints under the d_economy will also have a spatial impact on c_innovation. The d_economy is not constrained by geographical distance. Through adopting digital information technologies, it can facilitate the dissemination of new technologies and knowledge among regions, make up for the shortage of resource endowments in adjacent areas, and optimize the cooperation models and innovative business forms among regions (Zhang et al., 2023). The progression of the d_economy and the cross-temporal connections of various Internet platforms have accelerated the dissemination and application of c_innovation experiences and knowledge and can have spillover effects on surrounding areas.
Accordingly, this article puts forward Hypothesis 4: The d_economy exerts a spatial spillover impact on c_innovation.
With regard to the previously mentioned analysis and research hypotheses, this article’s theoretical hypothesis framework diagram is built, as shown in Figure 1.
Framework diagram of theoretical hypotheses.
Flowchart illustrating the influence of the digital economy on carbon-neutral technology innovation. It depicts direct and indirect effects. Direct effects include human capital, financing constraints, and industry-university-research cooperation. Indirect effects involve resource misallocation, subdivided through digital economy, digital industrialization, and industrial digitalization. Threshold effects feature Metcalfe's Law, scale economies, and collaborative innovation. Spatial effects cover flowing space, technological spillovers, and knowledge distribution. The overall focus is on improving resource allocation efficiency and breaking down regional and industry barriers.3 Model construction and variable measurement3.1 Model construction
To examine the four hypotheses put forward in this article, the direct, indirect, nonlinear, and spatial effects of d_economy on c_innovation are investigated by using a fixed-effect model, a mediating effect model, a dynamic threshold regression model, and a spatial Durbin model.
First, this article establishes a fixed-effect model to explore the direct implications of the d_economy and its two internal systems, namely digital_i and industrial_d on c_innovation:c_innovationit=α0+α1Dit+αnXit+λi+γt+εit
In Equation 1, c_innovationit denotes the explained variable, indicating the c_innovation of region i in period t. Dit is the core explanatory variable, including the level of regional d_economy (d_economyit), the level of digital_idigital_iit and the level of industrial_dindustrial_dit. Xit represents a set of control variables, including the financial development level (financeit), the industrial structure (industryit), the degree of government intervention (governmentit), and the degree of opening up to the outside world (openingit). α0 is the constant term. λi signifies the urban fixed effect, γt denotes the time fixed effect, and εit signifies the random disturbance term.
Second, this article introduces resource mismatch as an intermediate variable, and the mediating effect model is employed to probe the indirect impact mechanism of d_economy’s development level on c_innovation. Drawing on the research approach of Jiang (2022), the regression model below is built:resourceit=μ0+μ1d_economyit+μnXit+λi+γt+εit
In Equation 2, μ0 is a constant term, and the remaining variables correspond to those featured in Equation 1.
Furthermore, to explore the nonlinear relationship between d_economy and c_innovation, this article designs the following dynamic threshold regression model (take a single threshold, for example):c_innovationit=β0+β1c_innovationit−1+β2d_economyitId_economyit≤η1+β3d_economyitId_economyit>η1+βnXit+λi+γt+εitc_innovationit=θ0+θ1c_innovationit−1+θ2d_economyitIdigital_iit≤η2+θ3d_economyitIdigital_iit>η2+θnXit+λi+γt+εitc_innovationit=ω0+ω1c_innovationit−1+ω2d_economyitIindustrial_iit≤η3+ω3d_economyitIindustrial_dit>η3+ωnXit+λi+γt+εit
In Equations 3–5, d_economyit, digital_iit, and industrial_dit are threshold variables. η1,η2, and η3 are threshold values. I· is an indicative function. The remaining variables correspond to those featured in Equation 1.
Finally, the spatial Durbin model not only enables the introduction of spatial factors to reflect the spatial correlation of c_innovation but also tests the impact of other possible factors on c_innovation. Therefore, a spatial Durbin model is adopted in this study to explore the spatial impact of the d_economy on c_innovation, as specified below:c_innovationit=σ0+ρW*c_innovationit+φ1d_economyit+φ2Xit+φ3W*d_economyit+φnWXit+λi+γt+εit
In Equation 6W*c_innovationit denotes the spatial lag term of c_innovation. W signifies the inverse square matrix of spatial geographical distance. ρ represents the coefficient of spatial autoregression. φ1,φ2, φ3, and φn represent the estimated coefficients of each variable, and the remaining variables are the same as in Equation 1.
3.2 Measurement and description of variables3.2.1 Explained variable
c_innovation. Patent quantity provides a more accurate reflection of innovation level (Farbmacher et al., 2022) to comprehensively investigate the low-carbon technology, zero-carbon technology, and negative-carbon technology innovation in c_innovation. Among them, low-carbon technology innovation refers to technological innovations that reduce greenhouse gas emissions and lower energy consumption. Zero-carbon technology innovation involves developing and utilizing non-fossil energy to achieve nearly “zero” carbon dioxide emissions. Negative-carbon technology innovation is technological innovation for capturing, storing, and utilizing carbon dioxide. This research references the findings of Gong and Xiao (2024) and uses the logarithm of the number of invention applications for Y02 patents in the Cooperative Patent Classification (CPC) plus 1 as an indicator to measure c_innovation. The European Patent Office and the United States Patent and Trademark Office collaborated to create the patent categorization. It boasts advantages of unified standards, strong compatibility, and high subdivision. The categories of patents included in the CPC-Y02 patent classification system and their meanings are shown in Table 1.
CPC-Y02 patent categories and their meanings.
Id
Implication
Y02
Technologies or applications aimed at mitigating or adjusting to climate change
Y02A
Technologies for adjusting to evolving climatic conditions
Capturing, storing, isolating, or disposing of greenhouse gases
Y02D
Climate change mitigation technologies within the information and communication technologies sector
Y02E
Reduction of greenhouse gas emissions associated with energy generation, transmission, or distribution
Y02P
Technologies aimed at reducing climate impact during the production or processing of commodities
Y02T
Technologies for mitigating transport-related climate change
Y02W
Technologies for reducing climate impact in wastewater purification or waste control
3.2.2 Core explanatory variables
Following the National Bureau of Statistics (2021) taxonomy of the d_economy, this article gauges the d_economy’s development level across two dimensions: digital_i and industrial_d, in accordance with its conceptual definition and accounting for data accessibility at the city level. digital_i acts as the foundation for the d_economy’s development, encompassing industries such as software and information technology services, and telecommunications. Therefore, from the standpoint of digital industry development, digital_i is measured from two dimensions: scale and development status, as well as innovation capabilities. industrial_d refers to the integration of digital technologies and the real economy. Hence, it is defined from the viewpoint of integrating digital technology with the industrial sector, covering three levels: the primary industry, the secondary industry, and the tertiary industry. Table 2 details the evaluation index system of the d_economy. Considering that the selected indicators are multi-dimensional, this article employs the projection pursuit method optimized by the accelerated genetic algorithm (RAGA-PP) to compute the d_economy’s development level (Niu and Liu, 2021). This method effectively reduces multi-dimensional data to a low-dimensional space by optimizing the projection direction ajt, ensuring that the structural features and key information of the original data are retained as much as possible during the dimensionality reduction process. By carrying out a global search for the optimal projection direction, the numerical value of the optimal projection direction represents the weight. When the projection index function Qa reaches its optimal value, the one-dimensional optimal projection value zit of the d_economy can be obtained. The specific steps for calculation are as follows:
1. Standardized sample indicators
Evaluation index system of China’s regional d_economy development.
Primary indicator
Secondary indicator
Third-level indicators
Unit
Indicator attribute
digital_i
Scale and development status
The quantity of personnel in computing, service, and software businesses
Person
+
Per capita total telecommunications volume
Ten thousand yuan
+
Innovation ability
R&D spending of industrial enterprises beyond the designated scale
Ten thousand yuan
+
The quantity of patent applications in the d_economy industry
Item
+
industrial_d
Primary industry
The primary industry’s value added
Hundred million yuan
+
Secondary industry
The secondary industry’s value added
Hundred million yuan
+
Tertiary industry
The tertiary industry’s value added
Hundred million yuan
+
E-commerce transaction volume
Ten thousand yuan
+
Digital financial inclusion index
-
+
For positive indicators:X+i,jt=xi,jt−xminjtxmaxjt−xminjt
In Equation 7, xmaxj and xminj, respectively, signify the maximum and minimum values of the j variable. X+i,jt denotes the dimensionless data of the positive variable after normalization.
In Equation 8, zit represents the projected value of the d_economy index, and ajt stands for the unit projection direction of the j indicator. In Equation 9Qa represents the projection index function. In Equation 10Ez denotes the average value of zit, Sz is the standard deviation of zit, in Equation 11Dz is the local density of zit, and R represents the radius of the local density window, while ri,j indicates the distance between samples. ri,j=zi−zj. ut represents a unit step function, taking the value of 1 when t ≥ 0 and 0 when t < 0.
3. Refine the projection index function
maxQat=SZDZs.t.∑j=19a2jt=1
In Equation 12, maxQat represents the maximization of the objective function.
4. Calculate the d_economy index
Through step (3), the optimal projection direction value aj is obtained and placed in the projection function to calculate the projection values zit of each indicator, which is the d_economy index value.
3.2.3 Threshold variables
d_economy (d_economyit), digital_i (digital_iit), and industrial_d (industrial_dit) are the threshold variables. Among them, both digital_i and industrial_d utilize the projection pursuit method optimized by genetic algorithms for normalization processing.
3.2.4 Intermediate variable
Optimal resource allocation refers to a state where the free movement of factors leads to maximized social output within a market mechanism, while resource mismatch or market distortions signify deviations from this optimal state. In this article, resource mismatch is selected as the intermediate variable. With reference to the relevant studies of Hsieh and Klenow (2009), this article employs the production function to gauge the level of factor market distortion in urban areas. The extent of resource mismatch in each city is assessed by comparing the market distortion level of that city with the highest distortion level observed among all cities in the current year. The C-D production function is constructed, and the logarithm is taken, as follows:LnYit=c+aLnKit+bLnLit+εitdistKit=|aYit/ritKit−1distLit=|bYit/ditKit−1
In Equation 13, aYit/Kit and bYit/Kit represent marginal output of capital and the marginal output of labor, respectively. In Equation 14, distKit and distLit stand for the levels of capital and labor distortion. By combining the distortions in capital and labor, the overall market distortion degree is distit=distKitaa+bdistLitba+b, where Y is calculated using the gross regional product. K denotes the capital stock, which is assessed through the perpetual inventory approach. L represents the workforce count, indicated by the count of employment at the city’s end of the year. r is the capital price, set at 10%, representing a 5% depreciation rate and a 5% effective interest rate. d denotes labor expenses, reflected through the average salary of the people employed in each city in the current year. a represents the output elasticity of capital, and b indicates that of labor.
3.2.5 Control variables
To achieve a more precise and thorough insight into how the d_economy influences c_innovation, this article draws on existing studies and introduces the following control variables (Dian et al., 2024; Li and Yue, 2024; Huang et al., 2023; Wu et al., 2019): (1) Financial development level (finance): The financial development level is indicated by the ratio of year-end deposit and loan balances from financial institutions to the gross regional product. (2) Industrial structure (industry): The industrial structure can be depicted by the ratio of tertiary industry value added to the gross regional product. (3) Degree of government intervention (government): The ratio of governmental spending to gross regional product is employed as a measure for assessing the degree of government intervention. (4) Degree of opening up to the outside world (opening): The proportion of total imports and exports to gross regional product is used as an indicator of the degree of opening up to the outside world.
3.3 Sources of data and descriptive statistics
During the 12th Five-Year Plan period, China started emphasizing the growth of the information technology sector and cultivating it as a strategic emerging industry. Although the term “d_economy” has not been explicitly proposed, the extensive use of information technology and the strengthening of digital trends have established a firm basis for the subsequent advancement of the d_economy. The period from 2011 to 2022 witnessed the whole process of the initial rise of the d_economy to the in-depth development, and it is also a key period for the proposal and implementation of c_innovation. Therefore, this article selects 264 Chinese cities, ranging from 2011 to 2022, as the samples applied in its study. Logarithmic transformation is applied to the related variables to avoid heteroskedasticity and multicollinearity. Table 3 presents the descriptive statistics of each variable and the data sources.
Descriptive statistics of each variable.
Category
Variable
Mean
Sd
Min
Max
Data sources
Explained variable
c_innovation
4.4986
1.7554
0.0000
10.7158
incoPat Database
Explanatory variables
d_economy
1.2424
0.3356
0.2045
2.3529
China Urban Statistical Yearbook, China Science and Technology Statistical Yearbook, etc.
digital_i
0.6092
0.1942
0.1717
1.5255
industrial_d
1.1297
0.2855
0.1233
2.0006
Control variable
finance
2.5809
1.2427
0.5879
21.3018
industry
0.4297
0.1029
0.1015
0.8387
government
0.1916
0.0933
0.0439
0.9155
opening
0.1849
0.2888
0.0001
2.4913
The digital inclusive finance index data are sourced from the Digital Inclusive Finance Index of Peking University.
4 Empirical test and result analysis4.1 Collinearity test
Given that potential multicollinearity may exist among different variables, the variance inflation factor (VIF) is first employed to perform a collinearity test prior to conducting the baseline regression analysis. The outcomes in Table 4 reveal that each VIF value is strictly less than 5, which signifies no multicollinearity among the variables (Batrancea and Tulai, 2022).
Results of the multicollinearity test.
Variable
d_economy
finance
industry
government
opening
VIF
2.4700
1.9100
2.4900
1.7000
1.2000
1/VIF
0.4045
0.5245
0.4018
0.5885
0.8309
4.2 Analysis of benchmark regression results
Table 5 exhibits the benchmark regression outcomes on how the d_economy facilitates c_innovation. Among them, columns (1) to (4) show the estimation results with control variables added incrementally, with column (5) providing the estimation result incorporating the full set of controls. Note that the number of added control variables does not affect the positive impact of d_economy on c_innovation, and with the increase of the quantity of incorporated control variables, the enhancing effect of d_economy on c_innovation is generally increased. A 1% rise in the d_economy results in a 3.9918% growth in the c_innovation. This verifies that Hypothesis 1 is valid. Relying on the extensive application of technologies such as big data, cloud computing, blockchain, and artificial intelligence, the d_economy can reduce the difficulty and expense of obtaining information, accelerate the spillover of knowledge and technology, and facilitate regional c_innovation and R&D, thus overall boosting the c_innovation level.
Regression results of d_economy on c_innovation.
Variable
(1)
(2)
(3)
(4)
(5)
d_economy
3.4676***
3.6588***
3.6313***
3.9946***
3.9918***
(0.1622)
(0.1721)
(0.1721)
(0.1823)
(0.1823)
finance
0.0457***
0.0526***
0.0290**
0.0287**
(0.0139)
(0.0140)
(0.0145)
(0.0145)
industry
−0.6374**
−0.7321***
−0.7205***
(0.2089)
(0.2084)
(0.2093)
government
1.4698***
1.4753***
(0.2538)
(0.2540)
opening
−0.0492
(0.0808)
_cons
−0.0297
−0.3037*
−0.0670
−0.5760**
−0.5669**
(0.1569)
(0.1772)
(0.1930)
(0.2111)
(0.2117)
Time fixed
Yes
Yes
Yes
Yes
Yes
Urban fixed
Yes
Yes
Yes
Yes
Yes
N
3,168
3,168
3,168
3,168
3,168
R2
0.8263
0.8270
0.8275
0.8295
0.8295
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
Concerning control variables, the financial development level can markedly promote the improvement of c_innovation, which indicates that regions with a higher financial development level can offer adequate funding for the research of c_innovation projects. The coefficient of government intervention’s effect is positive and relatively notable, which shows that the government can establish and improve industrial policies related to carbon-neutral technology, guide traditional industries to transform toward a green and low-carbon direction, and provide certain support for c_innovation. However, the industrial structure exerts a negative effect on c_innovation. This result aligns with the findings reached by Huang et al. (2023). The possible cause may be that the unreasonable industrial structure makes it difficult to efficiently focus innovation resources in the field of c_innovation, and more resources such as capital, talent, and technology will flow to traditional high-carbon industries or other non-key fields, leading to inadequate resources for c_innovation research. There are certain restrictions on its development.
In addition, to further explore how the two systems of the d_economy affect c_innovation, this article adds the core explanatory variables of digital_i, industrial_d, and control variables into the full-sample baseline regression analysis. As illustrated in Table 6, column (1) and column (3) display the estimation findings with no control variables included, whereas columns (2) and (4) show the estimation outcomes after including the control variables. One can observe that, whether or not control variables are incorporated, digital_i and industrial_d both impose a notable enhancing effect on c_innovation, and the promoting effect of digital_i is stronger. This finding aligns with the Wang and Wei (2023) study on how digital_i and industrial_d affect enterprise innovation. Specifically, each 1% growth of digital_i will lead to a 4.9409% increment in c_innovation; for every 1% increase in industrial_d, c_innovation will increase by 0.9242%. The possible reasons are as follows: On the one hand, digital_i, as the foundational part of the d_economy, encompasses industries such as electronic information manufacturing, telecommunications, software, and information services. These are the industrial foundations for the development of the entire d_economy. Compared to traditional industries like industry, they inherently possess the advantages of being green and low-carbon. Furthermore, the digital industry can leverage the penetration and expansion of digital technologies to boost the upgrading of traditional industries, drive the transformation of industries towards intelligence and greenness, and lay a certain foundation for c_innovation. On the other hand, from the standpoint of the concept of industrial_d, integrating traditional industries and digital industries is a process that takes time and will not immediately lead to an increase in production efficiency. Corresponding environmental effects may also have a certain time lag. Thus, the promotion influence of industrial_d on c_innovation is relatively weak.
Regression results of digital_i and industrial_d on c_innovation.
Variable
(1)
(2)
(3)
(4)
digital_i
4.9772***
4.9409***
(0.1584)
(0.1589)
industrial_d
1.0459***
0.9242***
(0.1808)
(0.2154)
finance
−0.0244*
−0.0179
(0.0133)
(0.0158)
industry
−0.5478**
−0.8294**
(0.1957)
(0.2252)
government
−0.2451
0.0769
(0.2230)
(0.2828)
opening
0.0867
−0.1006
(0.0757)
(0.0869)
_cons
0.8198***
2.3912***
1.0990***
2.8300***
(0.0815)
(0.1562)
(0.1127)
(0.2256)
Time fixed
Yes
Yes
Yes
Yes
Urban fixed
Yes
Yes
Yes
Yes
N
3,168
3,168
3,168
3,168
R2
0.8501
0.8012
0.8511
0.8025
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
4.3 Endogeneity and robustness4.3.1 Endogeneity test
Consider the exclusion of important variables or the likelihood of a reverse causal link between the d_economy and c_innovation, leading to endogeneity issues. With reference to the research conducted by Chang et al. (2021), the topographic relief was selected as the instrumental variable for the d_economy. On one side, the topographic relief can function to reflect the complexity of a locality’s terrain, which in turn influences the installation and commissioning of digital infrastructure. In general terms, the larger the topographic relief, the higher the expense and difficulty associated with constructing digital infrastructure. Hence, the topographic relief fulfills the relevance condition for being used as an instrumental variable. On the other side, the topographic relief, as a natural factor, has no direct correlation with other economic variables and thus satisfies the exogeneity condition required for being used as an instrumental variable. Because the original data of the employed instrumental variable exists in cross-sectional form, following the approach of Nunn and Qian (2014), this article incorporates a variable that varies with time to build a panel instrumental variable. Therefore, an interaction term between the topographic relief and the time trend is created to serve as the test instrument variable. Using this foundation, the two-stage least squares method (2SLS) and the system GMM model are employed to conduct the model concurrently. The outcomes of these tests are presented in Table 7.
Results of endogeneity test.
Variable
(1)
(2)
(3)
One-stage regression
Two-stage regression
System GMM
L.c_innovation
0.5539***
(0.0086)
d_economy
3.9567***
2.2633***
(0.4087)
(0.0536)
iv
−0.0057***
(0.0007)
_cons
−0.7561***
−0.8411**
−0.6155***
(0.0198)
(0.3179)
(0.0616)
Control
Yes
Yes
Yes
Time fixed
Yes
Yes
Yes
Urban fixed
Yes
Yes
Yes
Cragg–Donald Wald F
67.3910
{16.3800}
Hansen
0.1440
AR (1)
0.0000
AR (2)
0.1520
N
3,168
3,168
2,904
The values within the square brackets are P-values. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
According to the outcomes of the two-stage least squares regression (1) and model (2), the first-stage regression demonstrates that the instrumental variables exhibit a significant correlation with the endogenous variable d_economy, which confirms the hypothesis regarding the instrumental variables’ correlation. Second-stage estimates indicate the coefficient of d_economy holds a notably positive value at the 1% level, confirming that the study’s finding holds after alleviating the endogenous problem. Moreover, the Cragg–Donald Wald statistic equals 67.3910, which exceeds the 10% critical threshold of 16.3800, thereby implying that there is no issue with weak instrumental variables. According to the results of the system GMM model (3), the AR test reveals that the model’s first-order sequences exhibit correlation, whereas the second-order sequences lack it, implying insignificant serial correlation in the original model’s error terms. It is worth noting that both types of models, after tackling the endogeneity problem, show that the d_economy significantly boosts c_innovation, which supports Hypothesis 1.
4.3.2 Robustness test
To verify the dependability of the findings, this study conducts a robustness test on the benchmark regression results through the following methods. (1) Lag the explanatory variable. With reference to Chang et al. (2025), this article chooses the lagged one-period d_economy as the core explanatory variable. A new regression is conducted on the d_economy’s empowerment of c_innovation, with the findings displayed in column (1) of Table 8. (2) Replace the explained variable. The number of published Cooperative Patent Classification (CPC) Y02 patents after adding 1 and then taking the logarithm, is used to measure c_innovation (Gong and Xiao, 2024), with the outcomes displayed in column (2) of Table 8. (3) Control for multi-dimensional fixed effects. Drawing on the method of Wu (2020), this article, based on the baseline regression, incorporates the interaction effect between provinces and time. The findings are exhibited in column (3) of Table 8. (4) Adjustment the research samples. Beijing, Tianjin, Shanghai, and Chongqing, which function as municipalities directly under the central government, have significant advantages in urban hierarchy, policy orientation, and economic size. Including these cities in the empirical sample might introduce a risk of bias to the results of the basic model test (Lu et al., 2025). After removing the four municipalities directly under the central government, the regression is conducted again, with the results displayed in column (4) of Table 8. It is observable that the d_economy’s coefficients under different testing methods are all significantly positive, confirming the robust stimulative influence of the d_economy on c_innovation.
Robustness test.
Variable
(1)
(2)
(3)
(4)
Lag the explanatory variable
Replace the explained variable
Control for multi-dimensional fixed effects
Adjustment the research samples
L.d_economy
2.3375***
(0.1959)
d_economy
3.1679***
3.8653***
4.0128***
(0.1699)
(0.2252)
(0.1842)
finance
−0.0065
0.0078
0.0030
0.0308**
(0.0151)
(0.0135)
(0.0150)
(0.0147)
industry
−0.7937***
−0.1507
−0.0193
−0.7128***
(0.2233)
(0.1950)
(0.2475)
(0.2107)
government
0.9711***
1.1944***
0.8233**
1.4954***
(0.2707)
(0.2366)
(0.2786)
(0.2563)
opening
−0.2351**
0.0420
0.0509
−0.0797
(0.0955)
(0.0753)
(0.0818)
(0.0844)
_cons
1.4040***
0.7910***
−0.4789
−0.6004**
(0.2232)
(0.1972)
0.3169
(0.2115)
Time fixed
Yes
Yes
Yes
Yes
Urban fixed
Yes
Yes
Yes
Yes
N
2,904
3,168
3,108
3,120
R2
0.7990
0.8619
0.9676
0.8287
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
4.4 Regional heterogeneity analysis4.4.1 Heterogeneity of geographical location
Because of discrepancies in cities’ geographic locations, urban infrastructure, economic development degrees, and government subsidies, the d_economy might exert a varied effect on c_innovation. In accordance with the categorization standards set by the National Bureau of Statistics, the sample cities are grouped into the eastern, central, and western regions to perform heterogeneity analysis. The outcomes in columns (1) to (3) of Table 9 reveal that the economic development level in the eastern, central, and western regions exhibits a notable positive impact on c_innovation, with the influence effect being western region > central region > eastern region. This research finding is basically consistent with the conclusion of Fan and Shen (2025). The primary cause might lie in the fact that, as a pioneer of China’s economic development, the eastern region has long been deeply engaged in the field of technology, has obvious advantages in initial technology endowment, has a high maturity of green innovation network, and has built a relatively dense relationship among various entities in the network. Although this perfect innovation network structure lays a solid foundation for technological innovation, it also limits the space for further development of core elements of the d_economy, such as information technology and big data. Given the relative stability of the existing network structure, there is limited room for the dividend release of c_innovation when new digital technology elements are integrated. In the midwestern areas, the green innovation network is still in its developmental phase, and the elements of the d_economy, such as information technology and big data, have broad application and integration space. When these elements are integrated into the local innovation system, they can be deeply integrated with local innovation resources, and then vigorously promote the promotion of c_innovation. This driving effect is not only reflected in the technical breakthrough level but also radiates to the optimization of the entire innovation ecology, releasing greater economic and environmental benefits. Especially in the western areas, traditional industries urgently need to be digitally transformed, and the plasticity of the green innovation network is strong, which makes the d_economy more significant in promoting c_innovation than the central region, showing great development potential.
Results of the regional heterogeneity test.
Variable
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Eastern region
Central region
Western region
Resource-based cities
Non-resource-based cities
Cities with a better foundation for innovation
Cities with a weaker foundation for innovation
d_economy
3.5866***
4.2540***
4.8317***
3.2450***
4.4680***
3.2831***
4.8618***
(0.3029)
(0.3141)
(0.3958)
(0.3158)
(0.2210)
(0.2307)
(0.2688)
_cons
−0.3883
−1.2458***
−0.9494**
−0.3178
−0.8210**
0.6189*
−1.6825***
(0.4106)
(0.3589)
(0.3794)
(0.3258)
(0.2711)
(0.3200)
(0.2618)
Control
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Time fixed
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Urban fixed
Yes
Yes
Yes
Yes
Yes
Yes
Yes
N
1,188
1,116
864
1,152
2,016
1,476
1,692
R2
0.8910
0.8256
0.7875
0.7836
0.8588
0.8920
0.7872
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
4.4.2 Heterogeneity of resource endowments
The economic growth of resource-intensive cities primarily relies on inputs like workforce and natural mineral resources, with most industries being led by the heavy chemical industry. These cities exhibit a low degree of technological advancement, particularly in the realm of green tech innovation and progression, which constitutes a weak link for resource-dependent cities. With reference to the National Sustainable Development Plan for Resource-based Cities (2013–2022), this article separates the sample into two groups: resource-based cities and non-resource-based cities in order to investigate whether d_economy can successfully foster the growth of c_innovation in resource-based cities. Finally, 96 resource-based cities and 168 non-resource-based cities are identified. Referring to the findings in columns (4) and (5) of Table 9, it is observable that under different levels of resource endowment, the d_economy’s development exerts a notable positive effect on c_innovation. The impact coefficient of resource-based cities’ d_economy development level on c_innovation stands at 3.2450, whereas the equivalent coefficient in non-resource-based cities is 4.4680. In contrast to resource-based cities, non-resource-based cities exhibit a stronger promoting effect, and this conclusion from the study aligns basically with the findings presented by Zheng et al. (2025). A possible cause could be that resource-based cities have a high degree of dependence on resources in their developmental processes. Their industrial structure is mainly composed of resource-oriented industries with high energy usage and high emissions (Kim and Lin, 2017). The abundant natural resources can bring them continuous income, which leads to the fact that the talents and funds needed for c_innovation are crowded out by the investment in resource exploitation. On the other side, resource-based cities have a low concentration of technology-based enterprises, and there is a shortage of technological resources. The resource industry sector is also a sector lacking technological progress and featuring weak demand for innovation. In consequence, resource-based cities lack the driving force for innovation, and the d_economy cannot fully exert its effects. Rather, resource-based cities tend to adapt to the needs of attaining the “carbon neutrality” goal. As the d_economy advances, they combine their own development advantages to promote the development of c_innovation.
4.4.3 Heterogeneity of innovation base
In response to environmental regulatory measures, enterprises, as key market participants, may reallocate innovation resources based on the severity of the policies (Takalo et al., 2021) to fulfill the demand for c_innovation. Hence, varying urban innovation foundations can result in diverse allocations of innovation resources, causing the heterogeneous effects of d_economy development levels on c_innovation. Based on the China City Innovation Index released by Fudan University’s Industrial Development Research Center to evaluate the innovation foundation levels among diverse cities, this article categorizes the research samples into 123 cities possessing a strong innovation foundation and 141 cities with a weak innovation foundation. Columns (6) and (7) in Table 9, respectively, present the regression outcomes for cities with strong and weak innovation foundations. The outcomes reveal that the advancement of d_economy in cities featuring strong and weak innovation bases alike exerts a notable positive impact on c_innovation. Specifically, the influence coefficient of the d_economy on c_innovation stands at 3.2831 for cities with a strong innovation foundation, while that of cities with a weaker innovation foundation is 4.8618. Cities with a weaker innovation base have a stronger catalytic effect than cities with a better innovation base. This article suggests that a plausible reason could be, based on the marginal effect theory, cities with a strong innovation foundation have already invested resources in c_innovation, achieved certain results in the early stage, and may face diminishing marginal returns when they continue to increase investment in d_economy. Cities with weak innovation foundation face diminishing marginal returns due to the small investment in the early stage. The investment of the d_economy can bring more obvious c_innovation results in a relatively brief period, and the marginal benefit is relatively high. The integration of d_economy can quickly fill the technical shortcomings and bring significant innovation.
5 Further analysis5.1 Indirect effect test
Table 10 displays the test outcomes for the mediating function of resource mismatch in the d_economy’s influence on c_innovation. As shown in Table 10, whether or not control variables are added, the d_economy’s coefficient holds a notably negative in both scenarios. Amid the division of global value chains, resource mismatch acts as a key factor that impedes the innovative development of enterprises. Currently, most scholars have confirmed that resource mismatch imposes a negative effect on the level of technological innovation (Wang and Guo, 2025). c_innovation projects are characterized by lengthy cycles, increased costs, and considerable risk. In the case of unreasonable resource allocation, investors usually prefer conventional projects with shorter return cycles and lower risks, which encroaches upon the funds required for carbon-neutral technology R&D, thereby hindering the improvement of c_innovation. d_economy is able to notably alleviate resource mismatch by accelerating capital flow, optimizing allocation, and providing diversified financing channels, directing more human, material, and financial resources to technology R&D projects that have real innovation potential and can effectively reduce carbon emissions, thereby fostering advancements in the level of c_innovation.
Test of the indirect effects of d_economy on c_innovation.
Variable
(1) c_innovation
(2) c_innovation
d_economy
−0.0356**
−0.0373**
(0.0140)
(0.0158)
_cons
0.1756***
0.2000***
(0.0136)
(0.0183)
Control
No
Yes
Time fixed
Yes
Yes
Urban fixed
Yes
Yes
N
3,168
3,168
R2
0.2650
0.2765
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
5.2 Threshold effect test
This article examines a dynamic threshold regression model where the thresholds are set as variables of d_economy, digital_i, and industrial_d to investigate the nonlinear effect of d_economy on c_innovation. First, an examination was conducted to verify the presence of panel threshold effects in the model and the number of thresholds. After 300 bootstrap samplings, the models constructed in Table 13 were subjected to single, double, and triple threshold tests (see Table 11). The outcomes show that the d_economy exhibits a single threshold, with the threshold figure being 9.9172. digital_i has a double threshold, namely, 0.4915 and 0.6237, and industrial_d has a single threshold, which is 0.8769 (see Table 12).
Results of the threshold effect significance test.
Threshold variable
Threshold
F-number
P-number
BS degree
Threshold
1%
5%
10%
d_economy
Single threshold
27.0900**
0.0133
300
28.7627
20.3024
17.3751
Double threshold
10.4300
0.3067
300
18.0967
15.8350
14.0493
digital_i
Single threshold
137.9000***
0.0000
300
27.7383
20.2854
16.4366
Double threshold
90.4000***
0.0000
300
22.9630
19.1588
15.0944
Triple threshold
50.7500
0.1400
300
68.6582
60.5253
54.5363
industrial_d
Single threshold
62.1100**
0.0000
300
28.9530
21.0705
18.7886
Double threshold
−8.7500
0.9999
300
26.4016
20.4496
15.3154
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
The confidence intervals and threshold values.
Threshold value
Inspect
Threshold estimates
95% confidence interval
d_economy
Single threshold
0.9172
[0.9087, 0.9182]
digital_i
Single threshold
0.4915
[0.4886, 0.4920]
Double threshold
0.6237
[0.6227, 0.6243]
industrial_d
Single threshold
0.8769
[0.8729, 0.8816]
Table 13 displays the outcomes of the threshold regression. The regression outcomes with the d_economy acting as the threshold variable are presented in column (1). The findings demonstrate that the d_economy exerts a notable promoting effect on c_innovation and shows a “marginal increasing” characteristic. Specifically, when the d_economy’s development level falls below the threshold of 0.9172, it exerts a beneficial influence on c_innovation, with an impact coefficient of 2.0815. When the d_economy is above the threshold, its impact coefficient on c_innovation grows to 2.2324. This research result is consistent with the view of economies of scale theory. That is, as the d_economy’s scale grows continually, the cost advantages of the d_economy in technology R&D, talent attraction, and other aspects gradually emerge, and the promotion role of c_innovation gradually increases. From the regression findings reported in column (2), where digital_i acts as the threshold variable, it is noticeable that under the constraint of digital_i, the enabling influence of the d_economy on c_innovation also presents a “marginal increase” feature. The growth of digital_i can revolutionize the research paradigm of c_innovation, strengthen the willingness for c_innovation, reduce the transaction costs for c_innovation entities in accessing innovation resources, and improve the independent innovation capabilities of carbon technologies. As displayed in column (3) using industrial_d as the threshold variable, when industrial_d is smaller than 0.8769, the d_economy’s influence on c_innovation is significantly positive, with a coefficient of 1.0252, and rises to 1.0496 when the level of industrial_d is higher than 0.8769. With the improvement of industrial_d, an industrial_d ecosystem has gradually formed and continuously improved, even forming a digital ecosystem that cuts across industries and regional areas (Hou et al., 2025). This can reduce the innovation and R&D costs across regions, industries, and enterprises, optimize the distribution of various innovation factors and resources, and drive improvements in c_innovation.
Parameter estimation results of the dynamic threshold model.
Variable
(1)
(2)
(3)
L.c_innovation
0.5841***
0.4203***
0.7737***
(0.0028)
(0.0022)
(0.0014)
d_economyd_economy≤0.9172
2.0815***
(0.0169)
d_economyd_economy>0.9172
2.2324***
(0.0153)
d_economydigital_i≤0.4915
5.4992***
(0.0326)
d_economy0.4915<digital_i≤0.6237
5.9285***
(0.0300)
d_economydigital_i≤0.6237
5.9366***
(0.0250)
d_economyindustrial_d≤0.8769
1.0252***
(0.0113)
d_economyindustrial_d>0.8769
1.0496***
(0.0098)
_cons
−0.4042***
−0.4887***
0.2561***
(0.0103)
(0.0118)
(0.0107)
Control
Yes
Yes
Yes
Time fixed
Yes
Yes
Yes
Urban fixed
Yes
Yes
Yes
AR (1)
0.0000
0.0000
0.0000
AR (2)
0.2560
0.1090
0.4320
Hansen Test of Overid
0.6440
0.6640
0.9350
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
Further calculations reveal that the average level of China’s d_economy during the observation period is 1.2424, which fell within the optimal threshold range. This indicates that the d_economy is currently capable of effectively driving the improvement of c_innovation. The average level of digital_i is 0.6092, which is still within the second threshold range. The gap from the lower limit value of the optimal range 0.6237 is comparatively slight, which signifies that the current improvement in digital_i contributes to boosting the d_economy’s exertion of its enabling effectiveness on c_innovation. The average level of industrial_d is 1.1297, which is also within the optimal threshold range. This reveals that under the constraints of industrial_d, the d_economy is capable of effectively driving improvements in c_innovation. Therefore, promoting the advancement of the d_economy and its two major subsystems is highly beneficial for enhancing c_innovation. The promulgation of documents such as the d_economy Development Plan for the 14th Five-Year Plan and the Action Plan for Carbon Peak Before 2030 has provided policy guidance for continuously advancing the d_economy, digital_i, industrial_d, and the enhancement of c_innovation. In the future, the government ought to proactively encourage the growth of the d_economy, digital_i, and industrial_d to further stimulate its positive contribution to the improvement of c_innovation.
In conclusion, under the constraints of the d_economy, digital_i, and industrial_d, the d_economy’s role in promoting c_innovation has shown a “marginal increasing” impact, further verifying Hypotheses 1 and 2 of this article. This research’s result is mostly aligned with the conclusion of Wang et al. (2022), but it focuses on the nonlinear influence of the d_economy on green technology innovation, rather than c_innovation. The outcomes of this article reveal that the d_economy not only aids green technology innovation but also promotes c_innovation.
5.3 Spatial effect test
Table 14 presents the findings of the global Moran’I test for the levels of c_innovation and the d_economy development degree of individual cities spanning 2011 to 2022. Both c_innovation and d_economy exhibit positive spatial autocorrelation significant at the 1% level when using a geographic distance-based weight matrix, which points to a notable spatial correlation between d_economy and c_innovation across every city. The local Moran scatter plot in Figure 2 shows that the c_innovation and d_economy activities in various cities are mainly located in the first and third quadrants, presenting “high-high” type aggregation and “low-low” type aggregation characteristics and having strong spatial correlation. Hence, adopting spatial econometric models to carry out additional research is justified.
Spatial correlation test.
Year
c_innovation
d_economy
Moran’I
Z- value
Moran’I
Z- value
2011
0.2134***
11.2176
0.2332***
12.2435
2012
0.2142***
11.2600
0.2296***
12.0591
2013
0.2088***
10.9760
0.2255***
11.8432
2014
0.2252***
11.8188
0.2189***
11.5063
2015
0.2296***
12.0450
0.2234***
11.7394
2016
0.2537***
13.2852
0.2476***
12.9910
2017
0.2676***
14.0023
0.2487***
13.0451
2018
0.2717***
14.2179
0.2551***
13.3740
2019
0.2484***
13.0172
0.2598***
13.6185
2020
0.2592***
13.5721
0.2615***
13.7053
2021
0.2455***
12.8706
0.2527***
13.2555
2022
0.2521***
13.2084
0.2510***
13.1682
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.
Moran scatter plot of c_innovation and the d_economy(a) and (b).
Two Moran scatterplots show spatial autocorrelation. Plot (a) depicts innovation with Moran's I value of 0.2521, while plot (b) shows the economy with Moran's I value of 0.2510. Both have positively sloped trend lines, indicating positive spatial correlation. Points are distributed with slight scatter around the trend lines.
After passing the LM, Hausman Wald, and LR tests, this article finally selects the spatial Durbin model (SDM) based on both time and urban fixed effects. As the spatial lag terms of both the independent and the dependent variable are added to the outcomes derived from the spatial Durbin model analysis, solely taking into account the direct regression results will overlook the independent variable’s marginal influence on the dependent variable, leading to bias in the estimation results (Anselin, 2001). Drawing upon the research conducted by Lesage and Pace (2009), the effects of independent variables on dependent variables within the spatial Durbin model are segmented into direct, indirect, and comprehensive effects. The direct effect here incorporates the cumulative effect of spatial feedback from a city’s spillover effect on adjacent cities, which is to say it includes the city’s own feedback effect and the spillover effect of its neighboring cities (Yuan et al., 2020). The indirect effect signifies the spillover impact, reflecting the indirect influence a city exerts on its neighboring cities. Total effect represents the summed value of these two types of effects in a city. The spatial Durbin model’s effect decomposition outcomes are provided in Table 15.
Estimation results of the spatial metrology model.
Variable
Direct effect
Indirect effect
Total effect
d_economy
4.1732***
2.2462***
6.4193***
(0.2033)
(1.1157)
(1.1049)
ρ
0.5892***
(0.0397)
σ2
0.1252***
(0.0032)
Control
Yes
Time fixed
Yes
Urban fixed
Yes
N
3,168
R2
0.7402
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
(1) Direct effect. The d_economy can significantly promote c_innovation. A 1% growth in d_economy is associated with a 4.1732% rise in the region’s c_innovation. (2) Indirect effect. The regression coefficient for the indirect effect is notably positive, signifying that the d_economy can exert a positive spatial spillover influence on c_innovation among geographically adjacent regions through spatial characteristics, thereby confirming Hypothesis 4. Likely reasons are that the d_economy can break through geographical barriers through information networks, enhance the movement and convergence of production factors across regions, promote the cross-regional dissemination of knowledge and technology, improve the learning and imitation efficiency of various market entities, and thereby increase the c_innovation level of surrounding cities. (3) Total effect. With the accumulation of positive direct and indirect effects, the d_economy exhibits a pronounced positive influence on c_innovation.
5.4 Spatial heterogeneity
Considering that the d_economy’s effect on urban c_innovation differs across spaces, this article groups and regresses each city with reference to the National Sustainable Development Plan for Resource-based Cities (2013–2020) and the digital infrastructure level of each city to examine how d_economy’s impact differs across city types. Considering geographical factors, this part is still based on the geographical distance weight matrix.
5.4.1 Heterogeneity of resource endowments
Following the National Sustainable Development Plan for Resource-based Cities (2013–2022), this article splits the samples into 96 resource-based cities and 168 non-resource-based cities. Results from the regression are shown in columns (1) and (2) of Table 16. In resource-based cities, the coefficients of both direct and indirect effects are markedly positive, signifying that the d_economy development has a notable role in driving c_innovation in the city and can also influence the improvement of c_innovation in neighboring regions via the spillover effect. In non-resource city regions, the d_economy exhibits statistically meaningful direct impacts on c_innovation; however, its spillover effects, although positive, fails to pass the significance test. This indicates that the spatial spillover effects of the d_economy have not been fully realized, possibly due to the “core city siphoning effect” masking the indirect effects.
Regression results of the spatial heterogeneity test.
Variable
(1)Resource-based cities
(2)Non-resource-based cities
(3)Cities with a low digital infrastructure level
(4)Cities with a high digital infrastructure level
Direct
3.6939***
4.4127***
3.6218***
5.3352***
(0.3454)
(0.2461)
(0.2444)
(0.3622)
Indirect
2.0322**
1.2302
0.0752
5.9601***
(1.0309)
(0.9620)
(1.1598)
(1.2449)
Total
5.7261***
5.6429***
3.6970***
11.2952***
(1.0256)
(0.9405)
(1.1297)
(1.2682)
ρ
0.2906***
0.4786***
0.5661***
0.1851***
(0.0582)
(0.0489)
(0.0457)
(0.0695)
σ2
0.1462***
0.1124***
0.1086***
0.1472***
(0.0061)
(0.0036)
(0.0034)
(0.0063)
Control
Yes
Yes
Yes
Yes
Time fixed
Yes
Yes
Yes
Yes
Urban fixed
Yes
Yes
Yes
Yes
N
1,152
2016
2076
1,092
R2
0.7309
0.7103
0.7896
0.7110
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. Standard errors are presented in parentheses.
5.4.2 Heterogeneity of digital infrastructure levels
Digital infrastructure constitutes the base for d_economy development and stands as a crucial impetus for the modernization of the ecological environment governance system and capabilities. Varied levels of digital infrastructure might exert an influence on the spatial spillover effects of the d_economy. To gauge digital infrastructure development, this research employs a set of metrics: the number of Internet broadband access users per hundred people, the number of mobile phone users per hundred inhabitants, and the density of long-distance optical cables (Fan and Shen, 2025). The entropy approach is adopted for carrying out the measurement. Taking the average digital infrastructure level in the sample observation period as a standard, cities are categorized into 173 with low and 96 with high digital infrastructure. The outcomes are exhibited in columns (3) and (4) of Table 16. For cities with either low or high digital infrastructure levels, the d_economy delivers a significant direct promoting effect on c_innovation. For cities with low digital infrastructure standards, the indirect effect is positive but fails to satisfy the significance test, while the indirect effect in cities with high digital infrastructure is markedly positive. The likely cause is that more advanced digital infrastructure, through a networked structure, can enhance the interconnection among industries and enterprises (Deng et al., 2023), optimize resource allocation, facilitate the breaking of geographical distance constraints between regions, promote better spatial resource allocation of c_innovation factors, and also help change the traditional innovation model of the c_innovation entities in this city, forming resource aggregation and scale effects, and further empowering the development of c_innovation.
6 Conclusions and suggestions6.1 Research summary
This article adopts a digital empowerment perspective and leverages panel data from 264 Chinese prefecture-level cities from 2011 to 2022. It establishes an assessment index system for the d_economy level based on the two aspects of digital_i and industrial_d. It uses fixed-effects models, mediating models, dynamic threshold models, and spatial Durbin models to probe the impact of the d_economy and its two major systems of digital_i and industrial_d on c_innovation. The primary conclusions follow: (1) The d_economy delivers a notable positive effect on c_innovation, and this conclusion holds following the execution of multiple robustness tests. Concerning regional heterogeneity, the boosting influence of d_economy on c_innovation presents the characteristics of western region > central region > eastern region, non-resource-based cities > resource-based cities, and cities with a weaker innovation foundation > cities with a better innovation foundation. (2) The two subsystems of d_economy, digital_i and industrial_d, both play a marked positive role in c_innovation, and digital_i delivers a stronger promotional impact on c_innovation. (3) Under the constraints of the d_economy, digital_i, and industrial_d, d_economy has a nonlinear impact on c_innovation, and both exhibit the characteristics of “marginal increase.” That is, as the d_economy, digital_i and industrial_d keep advancing, their promoting effects on c_innovation gradually increase. (4) The d_economy can promote the advancement of c_innovation by addressing the issue of resource mismatch. (5) d_economy exerts a positive spatial spillover influence on c_innovation. It can both improve c_innovation in the local area and foster the growth of c_innovation in nearby regions.
Variable abbreviation list.
Variable
Abbreviation
Carbon-neutral technology innovation
c_innovation
Digital economy
d_economy
Digital industrialization
digital_i
Industrial digitalization
industrial_d
Resource mismatch
resource
Financial development level
finance
Degree of opening up to the outside world
opening
Degree of government intervention
government
Industrial structure
industry
Research and development
R&D
6.2 Policy suggestions
1. Based on regional development differences, heterogeneous governance strategies should be implemented. Previous studies have found that the d_economy has different promoting effects on c_innovation in different geographical locations, resource endowments, and levels of innovation foundation. Thus, the government ought to design tailored policies by referencing each city’s geographical position, resource endowment, and innovation foundation. For cities located in the central and western regions, resource-based cities, and those with a relatively weak foundation for innovation, endeavors should be devoted to accelerating the advancement of digital infrastructure and the digital_i process, providing the necessary material and technological foundation for a deeper integration of the d_economy and the real economy. Simultaneously, digital transformation will be implemented for key industries and key enterprises, starting from individual cases and gradually achieving the digital transition and industrial upgrading of the entire economy. This will facilitate the balanced development of d_economy and c_innovation and help attain the “carbon neutrality” goal. Cities in the eastern region, non-resource-based cities, and those with a strong foundation for innovation, given their digital infrastructure is comparatively advanced, should not only accelerate the development of digital_i but also further deepen the integration depth and breadth of the d_economy with the real economy and fully leverage the environmentally friendly advantages of the d_economy.
2. Drive the deep progression of digital_i. Previous studies have found that digital_i has a stronger promoting effect on c_innovation than industrial_d. Consequently, there is a necessity to further boost the progression of digital_i and better leverage its role in facilitating c_innovation. Reinforcing the supply of relevant technologies is necessary to advance the growth of the digital industry. This involves intensifying efforts in core technology R&D, building digital industrial clusters, and upgrading digital infrastructure. We should promote the establishment of new digital infrastructure, such as information network upgrades, cloud-network synergy optimization, and deep integration of computing and networks, and improve the basic institutional framework of the data element market, activate the value of data elements, and unleash the vitality of data elements. A national cluster of digital technology laboratories should be established, with the government taking the lead and leading enterprises serving as the core, to tackle key digital technologies and cutting-edge technologies, providing solid technical support for promoting c_innovation.
3. Strengthen cooperation and exchanges among cities, and promote coordinated development of cities. Previous studies have found that d_economy exerts a positive spatial spillover influence on c_innovation. Therefore, it is necessary to cultivate a digital economic development model that promotes cross-regional collaboration to enhance communication and cooperation among cities in order to alleviate the imbalance in regional progress. On the one hand, government departments should inspire enterprises inside and outside the region to build digital service platforms, jointly carry out technology development, actively share transformation experience, cooperate in digital projects, and form a coordinated development model for regional d_economy development by promoting cooperation among enterprises and linkage of industry associations. On the other hand, government departments should build an open policy environment and service system, formulate trans-regional d_economy development plans, clarify the positioning and development direction of each region, and form a coordinated development model of d_economy across time and space. When formulating management policies, it is important to give due consideration to the radiation and leading influence of high-level neighboring cities on the target city. We should also strengthen cooperation and exchanges with high-level regions, such as leveraging the radiation and leading role of cities with high development levels, like Shanghai and Nanjing, on other cities that are developing more slowly.
6.3 Deficiency and prospect
Although this article makes a certain supplement to the lack of relevant research on d_economy and c_innovation, it also offers a theoretical reference for the research on the impact of d_economy on c_innovation, albeit its limitations require further attention. (1) This study takes the panel data of Chinese cities as the research sample. Although it can provide certain references for enabling the d_economy to boost the level of c_innovation, it does not involve comparative studies in other regions. In the future, more representative economic belts or economic circles like the Yangtze River Economic Belt or the Beijing–Tianjin–Hebei region can be selected as samples for empirical research to strengthen the practical value of the research outcomes. (2) The d_economy measurement index system built around 264 cities in China and the measurement model built are designed for the research samples of this article. The conclusion must be further verified by more empirical data.
Data availability statement
The datasets presented in this article are not readily available because the disclosure of the materials analyzed during the current study is subject to the restrictions under an ongoing project. The corresponding author is willing to share the datasets upon any reasonable request under necessary confidentiality agreements. Requests to access the datasets should be directed to Yiming Chen c55332025@163.com.
Author contributions
YG: Formal Analysis, Investigation, Methodology, Supervision, Writing – original draft, Writing – review and editing. CL: Conceptualization, Data curation, Funding acquisition, Methodology, Resources, Validation, Visualization, Writing – review and editing. YC: Conceptualization, Methodology, Software, Supervision, Writing – review and editing.
Funding
The author(s) declare that no financial support was received for the research and/or publication of this article.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declare that no Generative AI was used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
ReferencesAnselinL. (2001). Spatial effects in econometric practice in environmental and resource economics. Amer. J. Agr. Econ.83 (3), 705–710. 10.1111/0002-9092.00194BatranceaL. M.TulaiH. (2022). Thriving or surviving in the energy industry: lessons on energy production from the European economies. Energies15, 8532. 10.3390/en15228532BowmanJ. P. (1996). The digital economy: Promise and peril in the age of networked intelligence. Educom. Rev.10, 69–71. 10.5465/ame.1996.19198671CaoS.NieL.SunH.SunW.Taghizadeh-HesaryF. (2021). Digital finance, green technological innovation and energy-environmental performance: evidence from China’s regional economies. J. Clean. Prod.327, 129458. 10.1016/j.jclepro.2021.129458ChangT.XuY.YuH.WuH.XueY. (2021). What is the role of telecommunications infrastructure construction in green technology innovation? A firm-level analysis for China. Energy Econ.103, 105576. 10.1016/j.eneco.2021.105576ChangX.FuK.ContehM. (2025). How digitalization impacts the high-quality development of the manufacturing industry: evidence from China. Sustainability17, 5586. 10.3390/su17125586ChenY. (2020). Improving market performance in the digital economy. China Econ. Rev.62, 101482. 10.1016/j.chieco.2020.10148238620214DengX.GuoM.LiuY. (2023). Digital economy development and the urban-rural income gap: evidence from Chinese cities. PLOS ONE18 (2), e0280225. 10.1371/journal.pone.028022536854034DianJ.SongT.LiL. (2024). Facilitating or inhibiting? Spatial effects of the digital economy affecting urban green technology innovation. Energy Econ.129, 107223. 10.1016/j.eneco.2023.107223FanY.ShenX. (2025). Research regarding the impact mechanism of digital economy development on economic resilience—Mediating effect based on upgraded industries. Sustainability17, 5749. 10.3390/su17135749FarbmacherH.HuberM.LafférsL.LangenH.SpindlerM. (2022). Causal mediation analysis with double machine learning. Econ. J.25 (2), 277–300. 10.1093/ectj/utac003GhasemaghaeiM.CalicG. (2020). Assessing the impact of big data on firm innovation performance: big data is not always better data. J. Bus. Res.108, 147–162. 10.1016/j.jbusres.2019.09.062GongY.XiaoY. (2024). Impact of sci-tech finance policy on carbon neutrality technology innovation. China Popul. Resour. Environ.34 (10), 66–79. 10.12062/cpre.20240719HallH. B.Moncada-Paternò-CastelloP.MontresorS.VezzaniS. (2016). Financing constraints, R&D investments, and innovative performances: new empirical evidence at the firm level for Europe. Econ. Innov. New Tech.25 (3), 183–196. 10.1080/10438599.2015.1076194HanD.LuK.LiuP. (2025). Multiplication of number and quality rise: data elements enable low-carbon innovation. Humanit. Soc. Sci. Commun.12 (1), 810–817. 10.1057/s41599-025-05194-zHanJ.ZhengW.CaiH.MingQ.YanX.FuY. (2025). A study of the spatial spillover effects and mechanisms of the digital economy on tourism carbon neutrality ability. Environ. Dev. Sustain.10.1007/s10668-025-06293-2HsiehC.KlenowJ. P. (2009). Misallocation and manufacturing TFP in China and India. Q. J. Econ.124 (4), 1403–1448. 10.1162/qjec.2009.124.4.1403HouJ.LiH.FuY. (2025). Study on the nonlinear mechanisms and spatial effects of the digital economy on urban carbon emissions in the Yangtze River Delta region. Ecol. Economy41 (07), 30–40+48.HuangX.ZhouJ.ZhouY. (2022). Digital economy’s spatial implications on urban innovation and its threshold: evidence from China. Complexity2022, 3436741–25. 10.1155/2022/3436741HuangX.ZhangS.ZhangJ.YangK. (2023). Research on the impact of digital economy on regional green technology innovation: moderating effect of digital talent aggregation. Environ. Sci. Pollut. Res. Int.30 (29), 74409–74425. 10.1007/s11356-023-27279-537209350JiangT. (2022). Mediating effects and moderating effects in causal inference. China Ind. Econ. (5), 100–120. 10.19581/j.cnki.ciejournal.2022.05.00535154431JinF.QiZ.LiangY. (2021). Big data, financial industry agglomeration and green technological innovation. Rev. Econ. Manage.37 (04), 97–112. 10.13962/j.cnki.37-1486/f.2021.04.008KimD.LinS. (2017). Natural resources and economic development: new panel evidence. Environ. Resour. Econ.66, 363–391. 10.1007/s10640-015-9954-5LaneN. (1999). Advancing the digital economy into the 21st century. Inf. Syst. Front.1, 317–320. 10.1023/A:1010010630396LesageJ. P.PaceR. K. (2009). Introduction to spatial econometrics. Boca Raton, FL: CRC Press.LiX.YueS. (2024). Blessing or curse? The role of digital technology innovation in carbon emission efficiency. J. Environ. Econ. Manage.365, 121579. 10.1016/j.jenvman.2024.12157938936018LingS.GaoH.YuanD. (2024). Catalytic role of the digital economy in fostering corporate green technology innovation: a mechanism for sustainability transformation in China. Econ. Anal. Pol.84, 278–292. 10.1016/j.eap.2024.09.005LuY.ZhuangJ.YangC.LiL.KongM. (2025). How the digital economy promotes urban-rural integration through optimizing factor allocation: theoretical mechanisms and evidence from China. Front. Sustain. Food Syst.9, 1494247. 10.3389/fsufs.2025.1494247LuoS.YimamuN.LiY.WuH.IrfanM.HaoY. (2023). Digitalization and sustainable development: how could digital economy development improve green innovation in China?Bus. Strat. Environ.32 (4), 1847–1871. 10.1002/bse.3223LvT.GengC.FuS.ZhaoQ.HuH.ZhangX. (2025). Exploring the impact of the digital economy on carbon emission intensity: a case study of the urban agglomeration in the middle reaches of the yangtze river. Environ. Dev. Sustain.10.1007/s10668-025-06436-5NiuH.LiuZ. (2021). Measurement on carbon lock-in of China based on RAGA-PP model. Carbon Manage12 (5), 451–463. 10.1080/17583004.2021.1966513NunnN.QianN. (2014). US food aid and civil conflict. Am. Econ. Rev.104 (6), 1630–1666. 10.1257/aer.104.6.1630ShiY.ZhangT.JiangY. (2023). Digital economy, technological innovation and urban resilience. Sustainability15 (12), 9250. 10.3390/su15129250SongM.FisherR.KwohY. (2019). Technological challenges of green innovation and sustainable resource management with large scale data. Technol. Forecast. Soc. Change144 (144), 361–368. 10.1016/j.techfore.2018.07.055SongS.WenJ.LiY.LiL. (2024). How does digital economy affect green technological innovation in China? New evidence from the “Broadband China” policy. Econ. Anal. Pol.81, 1093–1112. 10.1016/j.eap.2024.01.008SturgeonT. J. (2019). Upgrading strategies for the digital economy. Glob. Strategy J.11 (1), 34–57. 10.1002/gsj.1364TanL.YangZ.IrfanM.Jian DingC.HuM.HuJ. (2024). Toward low-carbon sustainable development: exploring the impact of digital economy development and industrial restructuring. Bus. Strat. Environ.33, 2159–2172. 10.1002/bse.3584TakaloS. K.TooranlooH. S.PariziZ. S. (2021). Green innovation: a systematic literature review. J. Clean. Prod.279, 122474. 10.1016/j.jclepro.2020.122474TapscottD. (1996). The digital economy: promise and peril in the age of net-worked intelligence. New York: McGraw-Hill.WangH.GuoX. (2025). The impact of resource misallocation on green technology innovation: evidence from 288 cities in China. Humanit. Soc. Sci. Commun.12, 823. 10.1057/s41599-025-05225-9WangD.QiY. (2023). The spatial mechanism of the digital economy enabling dual circulation development: from the perspectives of digital industrialization and industrial digitalization. Foreign Econ. manage.45 (09), 3–21. 10.16538/j.cnki.fem.20221122.401WangL.ShaoJ. (2023). Digital economy, entrepreneurship and energy efficiency. Energy269, 126801. 10.1016/j.energy.2023.126801WangQ.WeiY. (2023). Research on the influence of digital economy on technological innovation: evidence from manufacturing enterprises in China. Sustainability15, 4995. 10.3390/su15064995WangC.LiuT.ZhuY.LinM.ChangW.WangX. (2022). Digital economy, environmental regulation and corporate green technology innovation: evidence from China. Int. J. Environ. Res. Public Health19 (21), 14084. 10.3390/ijerph19211408436360975WangL.ZhaoH.LiuJ.ZhuH.LiuY. (2024). How does the digital economy affect carbon emissions from tourism? Empirical evidence from China. J. Clean. Prod.469, 143175. 10.1016/j.jclepro.2024.143175WangY.LiuK.LiuF. (2024). Digital economy: a driver for smart carbon reduction. Environ. Dev. Sustain.10.1007/s10668-024-05916-4WuJ. (2020). A joint test for serial correlation and heteroscedasticity in fixed-T panel regression models with interactive effects. Econ. Lett.197, 109594. 10.1016/j.econlet.2020.109594WuH.HaoY.WengJ. (2019). How does energy consumption affect China’s urbanization? New evidence from dynamic threshold panel models. Energy Pol.127, 24–38. 10.1016/j.enpol.2018.11.057XieM.ZhangC. (2024). Can linguistic big data empower digital economy?: evidence from China. Econ. Anal. Pol.84, 1771–1787. 10.1016/j.eap.2024.11.007XuQ.ZhongM.LiuX. (2022). How does digitalization affect energy? International evidence. Energy Econ.107, 105879. 10.1016/j.eneco.2022.105879XueY.ChangT.WuH.LiuJ.HaoY. (2022). The emerging driving force of energy consumption in China: does digital economy development matter?Energy Policy165, 112997. 10.1016/j.enpol.2022.112997YangX.XuY.RazzaqA.WuD.CaoJ.RanQ. (2024). Roadmap to achieving sustainable development: does digital economy matter in industrial green transformation?Sustain. Dev.32, 2583–2599. 10.1002/sd.2781YuanR. (2025). The synergistic effect of digital economy and environmental governance on urban green development. Front. Environ. Sci.13, 1506794. 10.3389/fenvs.2025.1506794YuanH.FengY.LeeJ.LiuH.LiR. (2020). The spatial threshold effect and its regional boundary of financial agglomeration on green development: a case study in China. J. Clean. Prod.244, 118670. 10.1016/j.jclepro.2019.118670ZhangY.MaG.TianY.DongQ. (2023). Nonlinear effect of digital economy on urban-rural consumption gap: evidence from a dynamic panel threshold analysis. Sustainability15, 6880. 10.3390/su15086880ZhangY.YangS.LiZ.CaiM. (2025). Research on digital economy empowering urban low-carbon transition. Sustain. Environ.10 (3). 10.22158/se.v10n3p91ZhaoT.ZhangZ.LiangS. (2020). Digital economy, entrepreneurship, and high-quality economic development: empirical evidence from urban China. J. Manage. World36 (10), 65–76. 10.3969/j.issn.1002-5502.2020.10.006ZhengH.WangY.CaiJ. (2025). Bridging green innovation and energy efficiency: the digital economy’s unique role in carbon neutrality. Environ. Dev. Sustain.10.1007/s10668-025-06246-9Appendix